Resonant Switching Bench Power Supply?

[MWighton]

Hi all,

I am trying to decide if going for a resonant switching power supply will be a good idea vs a traditional hard switching supply.

Design Parameters:
Input Voltage: 48VDC
Output Voltage: 0-40VDC
Max Output Power: 1200-1400W

Note: Input voltage can change if someone has a good reason as to why they think it should.

Obviously I am using a buck topology, but again as per my note I am open to changing it.

Thanks for the help,
MWighton

[T3sl4co1l]

I don't think it really matters until you get up in voltage, and need isolation.  The motivation then being, it's difficult to make a transformer suitable for hard switching, due to the number of turns involved, and this neatly addresses the extra reactance (capacitance and leakage inductance).  Of course, just having better efficiency is a nice-to-have.

Or going to very high operating frequencies for compactness and efficiency purposes.

If you aren't very experienced with SMPSs, starting large, and starting with resonant, isn't a wise decision.  Small and conventional will show all the gotchas in SMPS design, and then you can scale up what you've learned.

Tim

[SiliconWizard]

If you haven't already, you could take a look at this:
http://www.ti.com/lit/an/slua159/slua159.pdf (oldish but gives the basics)
http://www.st.com/resource/en/application_note/cd00004440.pdf

TI has some integrated controllers.

But frankly, a 1K+ resonant buck converter sounds like quite an endeavour.

[jbb]

What’s your power electronics background like?

0-40 V DC

This could be surprisingly hard. Having a wide range of output voltages means having a wide range of duty cycles.

Many resonant switching cells only operate over a limited range of duty cycles, so you have a challenge there. I would suggest (given a bench supply isn’t super cost sensitive) that you have a look at auxiliary commutation circuits with an auxiliary MOSFET in. This gives you freedom to choose when to begin the resonant switching transition and thus adapt to wide changes in duty cycle.

Some resonant schemes also rely on a minimum load current to complete the commutation, which is a problem for a bench supply. Again, an auxiliary commutation circuit can help get around this.

Finally, output at very low voltage can be hard. This is because you need very small duty cycles and therefore very short on times. Soft switch commutations are slower than hard switches ones and therefor make low output voltage more difficult.

As soft switching components have losses, you won’t eliminate switching losses. But they will reduce a lot. We found that the switching losses dropped by about 75% from the original hard-switched approach.

[MWighton]

I currently work in the power supply division of a defense contractor. I am fairly new to power supplies and think doing a project like this would help me at my job. I was talking with a coworker and he also mentioned that resonant supplies will be hard to operate over large output voltage ranges. The reason why I was looking into resonant is because I went to a power conference held by TI and they spoke about them and I thought, "why not use that for my bench supply project?".

My other idea was to make multiple smaller supplies and current share them. They would be connected through a back plane.

Since I am still a noob with power supplies I think the best approach is to start with a smaller supply( 0-15V max power ~ 300-500) and then upscale from there. I have the tendency to want to jump right to the end goal without testing before hand... 

[jbb]

I currently work in the power supply division of a defense contractor. I am fairly new to power supplies and think doing a project like this would help me at my job.
...
I have the tendency to want to jump right to the end goal without testing before hand... 

It's great that you want to really get into your job like that.  You would certainly learn a lot by attempting a supply like this.  As it happens, I happened to have some LTSpice files for a similar PSU lying around.  The topic is pretty complicated and there are lots of gotchas, so it's probably best to start simple.

As T3sl4co1l points out, a classic hard-switching design will be reasonable with 48V input, and...

Small and conventional will show all the gotchas in SMPS design, and then you can scale up what you've learned.

If I might be so bold, maybe you should try to work your way up the complexity tree.  I assume that many of your employer's designs will use microprocessor control and suggest the following 'curriculum':

• Conventional 48V in (hard switched) buck with analog current control chip
• Conventional 48V in buck with microprocessor current control (need current sensing, gate drivers, firmware)
• Conventional 48V in multi-phase buck with microprocessor current control (current sharing, interleaving to reduce ripple current, phase dropping)
• 48V - 12V isolated resonant converter (e.g. active clamp flyback, active clamp forward or LLC)
• Auxiliary resonant buck

As an example of how complicated things can get for a wide range converter, I attached s a screen shot of a modelling attempt.  You can see that the circuitry is complex, requiring extra MOSFET, diodes, capacitors, coupling transformer and gate drive.  The timing is quite important and there are features like how much delay to impose between the main and auxiliary switch (to allow soft switching to happen, very dependent on load), the minimum on time of the auxiliary switch (got to get the energy out of La), the minimum on time of the main switch (implications for duty cycle) and the fact that at low load currents the dang thing doesn't turn off with soft switching. Oh, and all the timing changes when the DC input Vin changes :-(

Please note that the resonant inductor La is a bit too big and the resonant capacitors are WAY too big in this design, which causes slow switching. Also the turns ratio n should be more like 1.2 to 1.5 to get the energy out of La faster (and thus reduce minimum pulse width).  The problem is that once you fix these it becomes necessary to dynamically adjust the delay time between the auxiliary switch firing and the main switch firing (hello microprocessor control).

[Circlotron]

I currently work in the power supply division of a defense contractor. I am fairly new to power supplies

Since I am still a noob with power supplies <snip> I have the tendency to want to jump right to the end goal without testing before hand... 

Sorry, couldn't resist :-P

[MWighton]

Thanks so much for your input. I will try my best to restrain myself and follow your guide to working my way up to the resonant converter. Im sure I will come back at some point with questions/problems that I will need assistance on.